Individual missing sampling occasions

questions concerning analysis/theory using program MARK

Individual missing sampling occasions

Postby shqxuan » Mon Jun 05, 2023 3:07 pm

Hi all,
I am trying to use MARK to do the mark-recapture analysis with the CJS model to estimate the recapture rate. However, some individuals in my encounter histories have irregular missing sample occasions.
For example:
Code: Select all
.....100010.0
.....11100..1
.....110001.0
....011......
1...011101...
0...000111...
0010011100...
000001100....
1101110..0.0.
0000110100..0
1011111000.00
1101100000.00
011110000..0.

In chapter 2.3 of the MARK textbook, it says
dot’ is not an individual covariate...

However, when I used the encounter histories with the dots being individual covariates, it still generated the results without error or warning. What does MARK do here to deal with the encounter history? What can I do to deal with the analysis?

Similar questions were asked here:
viewtopic.php?f=1&t=4228&p=14014&hilit=missing+occasions#p14014
viewtopic.php?f=34&t=3926

Thank you!!
shqxuan
 
Posts: 3
Joined: Mon Jun 05, 2023 12:35 pm

Re: Individual missing sampling occasions

Postby cooch » Mon Jun 05, 2023 3:22 pm

MARK treats 'dots' in the encounter history as 'missing occasions', and (depending on the data type) accounts for things appropriate to the data type you're working with. Start by reading the -sidebar- starting on p. 24 of Chapter 4 of the MARK book:

http://www.phidot.org/software/mark/doc ... df#page=24
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: Individual missing sampling occasions

Postby shqxuan » Mon Jun 05, 2023 4:34 pm

cooch wrote:MARK treats 'dots' in the encounter history as 'missing occasions', and (depending on the data type) accounts for things appropriate to the data type you're working with. Start by reading the -sidebar- starting on p. 24 of Chapter 4 of the MARK book:

http://www.phidot.org/software/mark/doc ... df#page=24


Thank you so much for the reply.
My understanding is that 'missing a sampling occasion altogether' for all individuals in the encounter histories is the prerequisite to using 'dots'. While in my encounter history, there is no 'missing a sampling occasion altogether'.
To put it another way, does it still make sense to use CJS for my encounter histories?
If not, how do we interpret the results given by MARK with my encounter histories?

Really appreciate your patience!
shqxuan
 
Posts: 3
Joined: Mon Jun 05, 2023 12:35 pm

Re: Individual missing sampling occasions

Postby gwhite » Mon Jun 05, 2023 7:19 pm

So you're going to have to explain how you know an animal was missing while another was present. The one case I've seen that made some sense was trapping on 2 different study areas with the assumption that animals did not move between areas, and some occasions were not trapped on just one of the areas.

The real point is that if you know an animal is missing, then p = 0, which is what a dot does in an encounter history. But if you put in a zero, then how did you know it was present? A 1 in the encounter history is obvious -- the animal was present.

Gary
gwhite
 
Posts: 329
Joined: Fri May 16, 2003 9:05 am

Re: Individual missing sampling occasions

Postby shqxuan » Tue Jun 06, 2023 3:43 pm

gwhite wrote:So you're going to have to explain how you know an animal was missing while another was present. The one case I've seen that made some sense was trapping on 2 different study areas with the assumption that animals did not move between areas, and some occasions were not trapped on just one of the areas.

The real point is that if you know an animal is missing, then p = 0, which is what a dot does in an encounter history. But if you put in a zero, then how did you know it was present? A 1 in the encounter history is obvious -- the animal was present.

Gary

Sorry for not providing a clear explanation of our project and for not providing sufficient information, which led to confusion.

We are using MARK and the capture-mark-recapture in an unusual way as we are not looking at animals but the pathogenic bacteria in the babies. We are really interested in applying the capture-mark-recapture technique to our research to estimate the probability of failing to detect the presence of bacteria even when it might be there.

We are working on a longitudinal study in which we monitor the presence of bacteria in the babies by collecting samples from the babies at 13 consecutive time points across 2 years. Due to the sampling techniques, it is possible that people failed to collect bacteria while bacteria were present in the babies. We want to apply the capture-mark-recapture technique in our study to estimate the probability of failing to detect the presence of bacteria, especially with the CJS model where we can estimate the recapture rate.

What we currently do is that when we received the samples, we detected the presence of the bacteria. If bacteria are present in the sample, we labeled the sampling occasions as “1”, if not, we label them as “0”. However, some samples were never sent to us, so those missing sampling occasions were labeled as “dots”. The “dots” essentially represent “lost samples” and it is unknown if they were 1 or 0. Each encounter history represents an individual baby.

I would greatly appreciate any feedback or suggestions regarding the most suitable strategy.
shqxuan
 
Posts: 3
Joined: Mon Jun 05, 2023 12:35 pm

Re: Individual missing sampling occasions

Postby cooch » Tue Jun 06, 2023 3:51 pm

Then, you might what to consider looking at occupancy models (where the 'dot' problem Gary referred to isn't an issue for that data type). Occupancy analysis models variation in proportion of 'sites' occupied. Substitute 'patient' for 'site', and 'prevalence' for 'proportion occupied (by the pathogen)', and you're off.

You might want to have a look at the following papers to start:

http://canuck.dnr.cornell.edu/research/ ... -paper.pdf

https://besjournals.onlinelibrary.wiley ... 11.00156.x

https://digitalcommons.unl.edu/cgi/view ... m_usdanwrc
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University


Return to analysis help

Who is online

Users browsing this forum: No registered users and 7 guests

cron